Machine learning approach to an otoneurological classification problem

Abstract

In this paper we applied altogether 13 classification methods to otoneurological disease classification. The main point was to use Half-Against-Half (HAH) architecture in classification. HAH structure was used with Support Vector Machines (SVMs), k-Nearest Neighbour (k-NN) method and Naïve Bayes (NB) methods. Furthermore, Multinomial Logistic Regression (MNLR) was tested for the dataset. HAH-SVM with the linear kernel achieved clearly the best accuracy being 76.9% which was a good result with the dataset tested. From the other classification methods HAH-k-NN with cityblock metric, HAH-NB and MNLR methods achieved above 60% accuracy. Around 77% accuracy is a good result compared to previous researches with the same dataset.

Classification of patients and controls based on stabilogram signal data

“…More specifically, we tested the following methods: Classification tree (CT) (CART algorithm [4,5,14,25,45]), k-nearest neighbor classifier (k-NN) [10,15,45], Linear discriminant analysis (LDA) [2,8], Logistic regression (LR) [12,22], Least-Squares Support Vector Machine (LS-SVM) [38,39], Mahalanobis discriminant analysis (MDA) [6], naïve Bayes (NB) variants [33,34,45], Quadratic discriminant analysis (QDA) [8,21] and Random Forests (RF) [7,23]. These algorithms were selected because they have shown great performance in many applications [20,40,42,45] and they extend our earlied study [30] significantly. Naïve Bayes method was tested with and without kernel density estimation (KDE) [18].…”

“…Machine learning methods have been used earlier in inner ear diagnosis. For example, in [20,40,41,42] a set of otoneurological diseases (including Ménière's disease) were classified using machine learning methods successfully. More specifically, in [20] halfagainst-half architecture was used and applied with k-NN, Support Vector Machine (SVM) and naïve Bayes classifiers.…”

Machine Learning Techniques for Differential Diagnosis of Vertigo and Dizziness: A Review

“…Varpa K. et al [30] compared one-vs.-all and one-vs.-one methods in KNN and SVM and found that using multiple binary classifiers (one-vs.-one) improved the true positive rates of disease classes. Joutsijoki H. et al [65] used half-against-half (HAH) architecture with SVM, KNN, and naïve Bayes (NB) methods with HAH-SVM reaching similar accuracy as OVO-SVM. Juhola M. et al [66] tested the classification capability of neural networks on ONE database.…”

Cutting Edge Technologies in Otology Field

“…A machine learning system (Galactica) has been developed for discovering a new diagnostic rule from a patient database and the diagnostic accuracy of developed rules were over 80% [17,19]. One of the machines learning model, the support vector machine, had successfully recognized the differences between normal subjects and dizziness patients by assessing a sensor based vestibular measuring system [20].…”

“…In the otology field, attempts are being made to develop a novel algorithm to diagnose or classify various otoneurological diseases such as dizziness and tinnitus [17,18]. A machine learning system (Galactica) has been developed for discovering a new diagnostic rule from a patient database and the diagnostic accuracy of developed rules were over 80% [17,19].…”